CIN specific reporter tagging in hiPSCs amenable for genetic perturbations
To facilitate multiple rapid and efficient genome engineering in human iPS cells, we opted for a strategy to integrate doxycycline-inducible SpCas9 at the “safe harbor” AAVS1 locus in human IMR90 iPS cells by Homology Directed Repair (HDR) (Castano et al., 2017). To this end, we used the type II-B CRISPR–Cas9 system from Francisella novicida (FnCas9) which in our previous studies, showed a combination of high intrinsic target specificity and higher homology directed repair outcomes when compared with SpCas9 (Acharya et al., 2019). The donor cassette for Homology Directed Repair (HDR) contained components of TET-ON gene expression system for Cas9 with the rTetR (“reverse” Tetracycline repressor) activator driven by CAG promoter and the Tetracycline Response Element (TRE) promoter for SpCas9 (Supplementary Fig. 1a). A single guide RNA (sgRNA) was used to target FnCas9 to the intron 1 of PPP1R12C gene (AAVS1 site, (Mali et al., 2013)) (Fig. 1a, Methods). Puromycin antibiotic from the donor cassette allowed for the selection of iPSC clones with successful knock-in. Confirmation of genomic integration of SpCas9 in selected and propagated IMR90 hiPS cell clones was done by PCR using primers flanking the donor cassette and genome and a specific amplicon of desired size was observed in the case of knock-in clones only (Supplementary Fig. 1b-c). Finally, we confirmed the robust and tunable expression of SpCas9 upon doxycycline treatment after 48 hrs (Fig. 1b, methods) and the selected hiPSC clones also showed proper expression of pluripotency markers, OCT4 and SOX2 confirming successful generation of the iCas9 knock-in iPSC line (Supplementary Fig. 1d).
Next, we proceeded towards developing a cellular model for facilitating the identification, tracking and isolation of viable GABAergic neurons from the in vitro differentiated iPSC lines. Towards this, we attempted to integrate a fluorescent reporter that could specifically label GABAergic neurons in the iCas9 knock-in iPSCs. The enzyme glutamate decarboxylase 67 (GAD67) encoded by the GAD1 gene is responsible for synthesis of GABA inhibitory neurotransmitter as it catalyzes the production of gamma-aminobutyric acid from L-glutamic acid (Pinal & Tobin, 1998). We reasoned that tagging of GAD67, being a very early marker of the GABAergic fate with extensive expression in the soma, would enable inclusion of all diverse GABAergic interneuron subtypes encompassing those with unique developmental origins from different regions of the sub-pallium (Esumi et al., 2021). This is in contrast to previous studies where transcription factor NKX2.1 has been tagged for CIN migration assays defining only the medial ganglionic eminence (MGE) progenitors and derived GABAergic interneurons (Tamamaki et al., 2003; Yun et al., 2002). We constructed a donor DNA cassette with a P2A-mClover3 reporter and hygromycin antibiotic cassette under PGK promoter for gene targeting by doxycycline induction of SpCas9 in IMR90-iCas9 hiPS cells (Fig. 1a). The cassette was flanked by 800 bp homology arms against C-terminal of GAD1 gene for integration before the STOP codon. We confirmed the genomic knock-in after hygromycin treatment and selection of iPSC clones, by PCR genotyping (Supplementary Fig. 1e). To characterize the expression of transgenic mClover signal in human GABAergic cortical neurons, hiPSCs were differentiated to drive the development of CINs (Fig. 1c). Upon dual SMAD inhibition for neural induction, Purmorphamine, an agonist of Smoothened receptor of the SHH pathway was added for ventral patterning (Fig. 1c). By 3 weeks of differentiation, hiPSC-derived cells expressed Nestin, a neural stem cell marker and by 30 days, NKX2.1, indicating the cell identity of MGE progenitors (Supplementary Fig. 1f-g). Importantly, CINs were differentiated in culture and by day 35 began to show faithful mClover signal suggesting proper, development-specific expression of the reporter gene (Fig. 1d). We further established the specificity of the endogenous construct expression through co-immunostaining with GAD67 and NKX2.1 followed by quantitative image analysis (Fig. 1e). Strikingly, by day 50, 90% of GAD67 + cells were also positive for mClover in the tagged group (Fig. 1f) and a sustained expression of mClover was observed in the cultured CINs. Taken together, our iPSC derived reporter cell line captures all IN subsets and can be used for imaging-based studies.
ERBB4 −/− forebrain organoids display reduced neuronal projections
We wanted to study the role of ERBB4 in modulating the migration of human CINs. Hyder et al, report cases of individuals with chromosome 2q34 deletions affecting ERBB4, exhibiting intellectual disability in three families having genetic deletion of exon 2 (Hyder et al., 2021). In silico analysis of the outcome of exon 2 deletion shows the incorporation of a STOP codon in frame and results in a truncated protein of 40 amino acids. Such deletion might also prevent translation of the protein isoforms (Fig. 2a). Therefore, we constructed ERBB4−/− hiPSC line in the background of iCas9 GAD1-mClover by deletion of exon 2 (Wang et al 2015) (Supplementary Fig. 2a, methods). ERBB4−/− hiPS cell colonies were expanded and screened by PCR to identify pure knockout clones with genomic deletion of 436 bp confirmed by DNA sequencing (Fig. 2b, methods). The selected ERBB4−/− hiPSC clone was also characterized for pluripotency marker validation (Supplementary Fig. 2b-c). Finally, we differentiated the ECG-iPSCs into CINs and as previously observed for WT cells, these iPSCs expressed mClover by day 35 suggesting the knockout of ERBB4 did not affect IN formation (Supplementary Fig. 2d). Importantly, western blot from cell lysate upon differentiation of ERBB4−/− iCas9_GAD1-mClover iPSC (hereafter called ECG-iPSCs) into CINs (day 35) confirmed the knockout of ERBB4 protein (Fig. 2c). Importantly the GAD67 positive cells co-stained with ERBB4 only in WT with no ERBB4 expression in ERBB4−/−, confirming the specific knockout in GABAergic interneurons (Fig. 2d). Through these results, we show that human ERBB4 is not essential for GABAergic interneuron formation under in vitro conditions.
Upon successful knockout of ERBB4 in human CINs, we hypothesized that it might have a role in the migration of CINs, as speculated from mouse studies. Unlike mice, where stage specific IN migration can be examined using stained brain tissues, observing dynamic IN migration events in human brain sections is currently not feasible. To circumvent these limitations, we took help of a forebrain assembloid system by fusing two separately grown ventral and dorsal spheroids. It was expected that such a setup would facilitate the cortical interneuron migration from ventral to the dorsal forebrain and allow us to carefully dissect modulators of CIN migration (Bagley et al., 2017; Birey et al., 2022) (Fig. 2e).
To establish a system where migratory dynamics of CINs can be accurately quantified, we introduced variations in the assembloid culture system taking inspiration from other reported methods. To enhance the onset of neuroectodermal differentiation, we used an optimized guided approach of region-specific forebrain organoid formation that involved neuroectodermal Embryoid Bodies (EBs) induced by dual-SMAD inhibition along with Wnt activation for dorsalization (Chambers et al., 2009; Sloan et al., 2018). We followed a similar approach for ventral forebrain organoid induction where dual SMAD and Wnt signaling inhibition was coupled with SHH pathway activation. For dorsal identity, SHH antagonist, CyclopamineA was added to neural induction media, and ventral/striatal identity was performed as per previously established protocol using SHH agonist and WNT antagonists during neural induction (Bagley et al., 2017) (Fig. 2e; Supplementary Fig. 2e). Upon patterning and induction of the neuroectoderm, we observed subsequent formation of the neuroepithelium in stereotypical rosette-like structures (Lancaster & Knoblich, 2014) (Supplementary Fig. 2f).
We opted for the fusion of separately patterned organoids derived from hiPSCs on day 12 in culture by seeding them in 1.5ml MCT or U-bottom plates (Sloan et al., 2018) (Supplementary Fig. 2g). This modification from previous reports (Bagley et al., 2017) resulted in a robust number of successful fusions and also allowed for easy embedding in Matrigel after 2 days to promote growth and structural organization. Dorsal forebrain organoids (DFOs) and ventral forebrain organoids (VFOs) were also cultured separately towards differentiation. A specific mClover signal was observed only from the VFOs from day 30 onwards (Supplementary Fig. 2h). To confirm the successful derivation of the forebrain identity organoids, the expression of specific markers of the ventral forebrain GE subregions was validated. FOXG1, human forebrain marker was expressed in both dorsal and ventral forebrain while GSX2 (LGE and CGE marker) level increased in the VFOs, along with NKX2.1 and LHX6 (MGE markers). In contrast, PAX6 and TBR1 expressed only in the DFOs (Supplementary Fig. 2i).
Upon starting the 3D differentiation of ERBB4−/− hiPSC line, EBs were successfully formed by aggregation of hiPSCs and were patterned to the dorsal and ventral identity (Supplementary Fig. 2j). Separately generated VFOs at week 2 were integrated with DFOs in case of both WT iPSCs and ECG-iPSCs and after 30 days of differentiation, expression of mClover was observed only in the ventral identity regions in both WT and ERBB4−/− conditions (Fig. 2f). After 50 days in culture, mClover + GABAergic cells were also observed in unlabeled (dorsal) regions of the assembloids (Fig. 2g), confirming that these cells had migrated into the dorsal cortical regions, similar to the migrating CINs during human brain development (Bagley et al., 2017; Bajaj et al., 2021). Both control and ERBB4−/− assembloids exhibited expression of dorsal radial glia progenitors, PAX6 and postmitotic neuron marker, TBR1. By week 7, both assembloids (WT and KO) expressed markers of the GE progenitors, NKX2.1; interneuron marker, GAD67 and an interneuron subtype marker, Parvalbumin (PVALB). There was insignificant change in the expression of forebrain lineage markers in ERBB4−/− hiPSC-derived assembloids over control (Fig. 2h). In summary, the ECG-iPSCs derived assembloids exhibited the necessary hallmarks of human forebrain identity.
We observed that WT VFOs cultured in matrigel without their dorsal counterpart showed extensive neuronal projections upon neural differentiation from day 35 onwards. Strikingly, ERBB4−/− VFOs showed visibly reduced projections (Fig. 2i). Quantification of the surface area of the WT and KO organoids within the outer boundary at day 40 and day 60 revealed no difference in organoid size between the two although both organoids showed a substantial increase in the surface area from day 40 to day 60 as expected (Fig. 2j). Upon quantification of radial distance from the organoid rim, a significant decrease in the migration distance from the organoid periphery was observed in ERBB4 KO VFOs compared with the WT VFO group (Fig. 2k-l). The identity of the projections at the organoid periphery and presence of differentiated GABAergic neurons in ERBB4 KO, was checked by immunohistochemical staining of organoid sections which confirmed that the cells expressed GABAergic interneuron marker, GAD67 (Supplementary Fig. 2k). Thus, 3D organoid modeling displayed a lack of neuronal projections in human ERBB4−/− cortical interneurons.
Loss of ERBB4 impairs migratory properties of cortical interneurons
Next, we proceeded towards the dynamic live cell imaging of CINs in the fused assembloids. Traditionally, studies of GABAergic interneuron migration derived from 3D cultures rely on whole mount imaging or organotypic slice preparations of assembloids (Supplementary Fig. 3a, (Bajaj et al., 2021)). However, live intact assembloid imaging suffers from lower resolution due to lack of light penetration and an imaging depth of around 50–150 um only (Andersen et al., 2020; Meng et al., 2023). To determine whether our ECG-iPSC derived assembloids were amenable to live imaging, we evaluated two different methodologies. Firstly, we performed culture of organotypic slices of assembloids for time lapse imaging of migrating CINs (Supplementary Fig. 3b; Supplementary Movie 1). However, due to the fragile nature of vibratome processed cortical slice preparations from forebrain assembloids, detailed analysis of movements of cells within the slice could not be recapitulated. In addition, we faced technical difficulties associated with reproducible mechanical sectioning of soft samples like organoids and in discerning the proper orientation of fused forebrain regions. For our study, the latter was particularly critical since we specifically investigated CINS migrating towards the dorsal surface.
We then imaged the ECG-iPSCs derived assembloids using Light Sheet Microscopy (LSM), which has also been utilized to image 3D organoids but it is feasible for only small or clarified spheroid/organoid samples (Boutin et al., 2018; He et al., 2022). A persistent challenge in imaging inside live tissues is the reduced optical penetrance and low signal resolution. To this end, we experimented with optimizing signal:noise ratios through the supplementation of non-toxic Opti-prep or Iodixanol for refractive index tuning of surrounding culture media (Boothe et al., 2017). The advantage was majorly limited to relatively small-sized spheroid samples and failed with large and complicated organoids (Supplementary Fig. 3c; Supplementary Movie 2; Supplementary Table 1).
Since whole assembloid imaging wasn’t able to capture CIN migration effectively, we next proceeded to dissect these parameters with greater granularity at the level of individual VFOs. Upon observation of reduced neuronal projections around the periphery of ERBB4−/− VFOs, we addressed the question of whether these interneurons show reduced motility in in vitro migration assays. To ascertain if ERBB4 regulates migration of human CINs, a simplified two-dimensional neurosphere outgrowth migration assay was performed. More than one millimeter sized 42-days old WT and ERBB4−/− VFOs were physically dissociated into visibly smaller aggregates and plated on PLO-Laminin coated dishes/coverslips (Supplementary Fig. 3d-e). 3 days post-attachment extensive projections of mClover + interneurons were observed moving away from the periphery of the dense WT VFO aggregate (Fig. 3a). Interneurons exhibited reduced neuronal projections around the periphery of ERBB4−/− VFO (Supplementary Fig. 3f) similar to the previous observations in matrigel. Five days post-plating migrated cells from the organoids were divided into two groups defined by the distance of migration (Fig. 3b). For comparison across different aggregates, the radial distance from the perimeter of the organoid to the outer circumference of migrated cells was calculated, and the number of interneurons that migrated from the periphery to 100 µm distance as well as beyond 100 µm distance were quantified. The motility observed showed visible differences as in the case of the ERBB4−/− interneurons. Notably, individual long projections and branching were absent and cells displayed collective cell migration outside VFO periphery (Fig. 3c). Taken together, significantly fewer ERBB4−/− interneurons could migrate to 100 µm and beyond, indicating a deficit in migration.
Next using time lapse imaging, we determined the migratory dynamics of cortical interneuron migration in high resolution, we imaged migratory dynamics of cortical interneurons and their saltation parameters on coated coverslips. Post 26 hours, we observed an increase in the population of cells close to the VFO periphery in the same field due to the proliferation of cells from VFO (Supplementary Fig. 3g). However, the overall distance migrated individually by WT interneurons was found to be lower than that observed in shorter timescales during slice culture assays (Supplementary Movie 3). Therefore, migration from solitary VFOs on coverslips even for WT CINs was inefficient. This is in close agreement with a previous report where saltatory migration of human interneurons from solitary VFOs plated on coverslips was quantified to be inefficient or absent (Birey et al., 2017).From the observations of 2D CIN migration assays, it was evident that a combination of physical substrate-driven mechano-sensation along-with intra-cellular signaling and inter-cellular interactions are crucial for the tangential migration of GABAergic CINs. Additionally, assembloid sample preparations with physical sectioning result in tissue damage and artefacts. Approaches involving slice culture in combination with optical sectioning via Single Plane Illumination Imaging (SPIM) face imaging bottlenecks owing to the thick nature of the tissue, resulting in poor resolution and contrast. To overcome these limitations, we utilized suspended fiber nano-nets, that allow 3D organoids to directly interact with polystyrene fibers (Jana et al., 2019). One of the key motivations for studying interneuron migration on these nano-nets was to interrogate single cell migratory behavior and identify subtle changes in the dynamics of ERBB4−/− migrating cells. The diameter of the suspended fibers was kept at 500 nm which is closer to the physiological ECM fiber bundles (Revell et al., 2021; Siadat et al., 2021) (Supplementary Fig. 3h). Importantly, migrating neurons utilized the nanofibers as tracks, allowing them to migrate and separate out from the organoids. This enables characterizing migration dynamics in ERBB4−/− CIN at high resolution.
Different arrays of aligned nanonets were used to mimic the brain micro-fibrillary environment and investigate the migration of CINs from VFOs over 20 hrs (Supplementary Fig. 3i). Ventral forebrain organoids were dissociated and seeded on aligned fibers and phase contrast images were acquired every 10 min for 24 hours’ time period (Supplementary Movie 4). Extensive long neurite projections and exploration of the surrounding environment on was observed in WT interneurons Our findings match the previous reports of neurite extensions on aligned fiber networks (Bakhru et al., 2011). To validate the identities of these neurons emanating from the WT VFO across various scaffolds, samples were fixed post-imaging and immunostained for GAD67 marker. Migrating cells on the fibers expressed this interneuron-specific marker (Supplementary Fig. 3j). Interestingly, long-distance migration (beyond 500 µm) of interneurons was observed on aligned fibers from the WT VFO perimeter (Fig. 3e) within 3 days of plating. In contrast, the extent of migration was absent when experiments were previously done on flat coverslip substrates. For ERBB4−/− CINs, collective persistent migration with reduced individual projections was observed (Fig. 3d; Supplementary Movie 5).
To further explore the role of underlying fiber geometry, disassociated WT VFOs were seeded on crosshatch nanonets and extensive migration was observed. Interestingly, long-distance migration of interneurons was observed on fibers of the migration scaffolds, beyond 500 µm from the WT VFO perimeter (Fig. 3e) within 3 days of plating. Such an extent of migration was absent when experiments were previously done on flat coverslip substrates. Upon using parallel orientation nano-fibers, unidirectional helical locomotion of interneurons was observed in both WT and KO interneurons with cells attaining bipolar morphology (Supplementary Movie 6; Supplementary Fig. 3n-o). To characterize the whole range of migratory behaviors, phase images of WT VFO samples were acquired every 3–5 min interval and position of interneurons tracked (Fig. 3f). Interestingly, we observed the interneurons to attain multipolar morphology and show branching of the leading process, possibly for the exploration of the extracellular fibrous environment (Fig. 3g). Migration trajectories of individual interneurons was traced and analyzed for their characteristic saltation movement. A swelling ahead of the trailing part with the nucleus behind it was observed and within three hours of swelling formation, the nucleus made a forward translocation (Fig. 3g; Supplementary Movie 7). These discontinuous saltatory movements confirmed that the behavior of these migrating cells resembled the unique GABAergic interneuron migration dynamics (Bellion et al., 2005).
After 24 hrs. of imaging, it was observed that ERBB4−/− interneurons showed decreased migration ability outside the VFO (Supplementary Movie 8) and very few interneurons could escape out of the VFO along the fiber with saltatory motion (Supplementary Fig. 3l). Interestingly, few cells showed leading projections extending onto the fibers and a trailing nucleus but were unable to perform a single nuclear translocation event (Fig. 3h). ERBB4−/− VFOs fixed on scaffolds and immunostained for interneuron marker, GAD67, showed expression in neuron projections confined within the outer rim of the organoid (Supplementary Fig. 3k). Upon quantification of saltation parameters, it was found that the length of saltation was significantly increased for ERBB4−/− interneurons compared to the WT VFO (Fig. 3i). However, no significant difference was observed in the average migration rate of KO interneurons (Fig. 3j; 42.9 ± 4.8 µm/hr for WT; and 46.86 ± 6.5 µm/hr for KO; n = 16 cells). It has also previously been suggested that nucleokinetic speed is not a major determinant of interneuron migration due to the arrhythmic nature of motion (Nichols et al., 2008). We therefore counted the saltatory events during a defined period for both WT and ERBB4−/− CINs and found that the frequency of these nuclear translocations was reduced in the KO (Fig. 3k). Thus, the difference in mean saltation length between WT and ERBB4−/− interneurons was produced largely by increased duration of single step. Extension and retraction of branches and leading process oriented in all directions from the interneurons were observed while cells also exhibited switching from one fiber to another and changes in their direction of motion on fiber networks (Supplementary Fig. 3m, Supplementary Movie 9).
The reduced migration on fibers for ERBB4−/− CINs was striking, suggesting that ERBB4 is critical for migration of interneurons. The observed alteration of saltatory event duration resulting in nucleokinetic abnormalities could be a possible contributor to the pathological condition. Overall, our findings suggest that nanofibers are a suitable artificial ECM that can be a better mimic of physiological matrix to potentially improve screening of cell migration parameters.
ERBB4 regulates human CIN migration via RHOA/RAC1 signaling
To further address the exact role of ERBB4 and associated signaling pathways in human CIN migration, bulk transcriptomics of ventral:dorsal organoid fusions at day 60 time point was performed. This time point for study was chosen since mClover expressing cells were observed by day 35 in the assembloids in culture and also increased migration was observed from day 40–60 in control assembloids similar to previous reports (Bagley et al., 2017). Principal component analysis (PCA) performed on the whole transcriptome of assembloids at day 60 showed distinct separation of control versus knockout samples from different assembloid batches (Supplementary Fig. 4a) Out of the total differentially expressed genes, 3,868 genes were downregulated and 3,277 genes were upregulated in ERBB4−/− assembloids compared with controls (p value < 0.001; FDR < 0.01; Fig. 4a; Supplementary Data).
One of the ERBB4 membrane protein isoforms has a phosphotidylinositol-3 kinase (PI3K) binding site and is known to modulate the PI3K/Akt signaling pathway (Gambarotta et al., 2004; Kainulainen et al., 2000) while all isoforms can stimulate the MAPK pathway (Sundvall et al., 2008). In the transcriptomics data, differentially expressed genes were linked to the PI3K/AKT signaling pathway (PIK3C2A, PIK3R1, PIK3R3, AKT1S1, AKT2) and mTOR pathway (RHEB, TSC1). These analyses revealed that ERBB4−/− assembloids recapitulated known pathways associated with the protein in the developing human cortex (Supplementary Fig. 4b). Gene Ontology (GO) analysis found highly significant enrichment of terms for biological processes related to protein phosphorylation, regulation of actin and microtubule cytoskeleton organization, neuronal migration, dendrite morphogenesis and ER-Golgi mediated transport processes, which were all significantly downregulated (Fig. 4b-c, Supplementary Data). Similarly, significant upregulation of calcium ion transmembrane transport, cilium assembly, cell-cell adhesion and excitatory post-synaptic potential biological processes was also observed (Fig. 4d). Importantly, knockout assembloids revealed the downregulation of genes associated with axon guidance, mTOR pathway and regulation of actin cytoskeleton pathway suggesting changes in migratory dynamics. In summary, the transcriptional landscape of ERBB4−/− assembloids agreed with our observed migratory CIN dysfunction.
At the level of diseases, common gene analysis between the differentially expressed gene data and available risk genes for intellectual disability and epilepsy showed significant overlap (ID Fold Enrichment: 0.88 and p-value: 7.92e-10; Epilepsy Fold Enrichment: 0.82 and p-value: 2.36e-08; Fig. 4d). Furthermore, the DisGeNET database through the DAVID tool linked the DEGs to intellectual disability disorder (Supplementary Data 4). The expression level of PI3K/AKT pathway genes PIK3R1, RHEB, TSC1, AKT1S1 were validated to be downregulated using real-time quantitative PCR and RHOA/RAC pathway genes RHOA, RAC1, CDC42, ROCK, ACTR6, ARHGAP5 were also significantly downregulated in the ERBB4 KO assembloids (Supplementary Fig. 4c). There was no significant change observed in expression of TSC2 and PTEN. Interestingly, functional interaction of the 25 genes under neuron migration (GO:0001764) term that were downregulated in ERBB4−/− assembloids, using STRING (https://string-db.org/) highlighted a densely interconnected network of proteins centered on RHOA, NTRK2 (activator of RAC1), and FYN (Supplementary Fig. 4d, Supplementary Data 2). Thus, through a detailed analysis of transcriptional signatures from ERBB4−/− assembloids, we identified its functional impact in the dynamic expression of disease risk genes during human brain development modeling in vitro.
We took advantage of genetically engineered hiPSCs in which GAD1 was tagged with mClover fluorescent protein to generate forebrain assembloids for WT and ERBB4−/−. Cell specific reporter tagging allowed for FACS based purification of GABAergic CINS from day 60 WT and ERBB4−/− assembloids to decipher the ERBB4−/− CIN protein abundance differences (Fig. 4e). Dimensionality reduction through principal-component analysis (PCA) showed distinct molecular signatures (Supplementary Fig. 4e). A total of 2050 protein groups were identified with FDR ≤ 1% in the spectral ion library. The analysis of CIN proteome revealed sufficient peptide coverage for high-confidence quantitative analysis of 1411 proteins (FDR < 0.01) with at least single unique peptides across all samples. (Supplementary Data). A more supervised approach to explore specific proteomic differences between WT and ERBB4 KO defined 266 differentially abundant proteins (Fold change cut off = ± 1.5; p-value < 0.05). Further analysis revealed the significant differential expression of peptide fragments belonging to 134 proteins in ERBB4−/− KO CINs were upregulated and 132 proteins were downregulated (Fig. 4f).
Gene set enchichment analysis using Reactome (Gillespie et al., 2022) biological pathways on the differentially abundant proteins showed highly significant association with terms related to regulation of expression of SLITs and ROBOs, signaling by ROBO receptors, RHO GTPases activate PKNs and RHO GTPase effectors (Benjamini–Hochberg-corrected p value < = 4.58E-10 Fig. 4g; Supplementary Data 3). 42 differentially enriched proteins within the ERBB4−/− KO population were also involved in axon guidance pathways.